Few-Shot Learning for Rooftop Detection in Satellite Imagery

Deep Learning Tutorial

Giorgio Coppala, Nadine Daum, Elena Dreyer, Nico Reichardt

Problem Setting

  • Cities need accurate rooftop maps to plan and scale solar PV installations

  • Manual rooftop labeling is slow and costly

  • Every city looks different → traditional models do not generalize well

Idea:

  • Few-shot learning makes segmentation possible with only a handful of labeled examples

Dataset: Rooftops of Geneva

  • Satellite Images: High-resolution RGB satellite images of Geneva available on Huggingface
  • Size: 1,050 labeled image-mask pairs
  • Task: Binary segmentation masks (rooftop vs background)
  • Geographic splits: 3 grids/ neighborhoods (1301_11, 1301_13, 1301_31)
  • Image size: 250x250 pixels
  • Categories: Industrial, Residential

Few Shot Learning in General

tbd

Prototypical Network

(modified figure from paper) SRPNet

  • high-level schematic (support → prototype → similarity → segmentation)

  • 1-way-1 shot –> explain what it means

  • Data Preprocessing (e.g. Augementation, Geographic Splits)

  • Model Architecture (feature Exctraction, CNN –> Number of Layers, Backbone)

  • Training strategy

    • Loss function

    • Evaluation metrics

(Preliminary) Results

  • Show performance for 1-shot / 5-shot / full-data comparison

  • Show predicted masks

Wrap-Up/ Discussion

GitHub Repo

What we still need to finalize:

  • insert bullet point here

  • insert bullet point here

Questions to discuss in class/ lynn

  • strengths

  • weaknesses

  • failure cases (shadows, tiny rooftops)